[CCCL] Add missing CCCL components: c2h, nvbench_helper, cmake, cudax, AGENTS.md

Added 863 files from NVIDIA/cccl sparse checkout:
- c2h/ (27 files): Catch2 test helpers — generators, validators, runner
- nvbench_helper/ (10 files): Benchmark harness utilities
- cmake/ (29 files): CMake presets and build helpers
- cudax/ (794 files): Experimental CUDA extensions
- AGENTS.md: NVIDIA's official AI agent instructions for CCCL
- CMakePresets.json: Standardized build configurations
- cccl-version.json: Version tracking

Also added CCCL_ASSET_MAP.md mapping all 4295 CCCL files to
competition value and PRD items.

cccl_upstream now covers 100% of competition-critical assets:
- 27 tuning headers (SM80/90/100 benchmark data)
- 32 dispatch headers (algorithm implementations)
- 60 Thrust examples (correctness verification)
- 217 CUB Catch2 tests (regression matrix)
- 153 CUB benchmarks (parameter space search)
- 18 CUB examples (API verification)
- 27 test helpers + benchmark harness
- 794 cudax experimental extensions

cccl_upstream/cudax/test/green_context/green_ctx_smoke.cu

@@ -0,0 +1,107 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of CUDA Experimental in CUDA C++ Core Libraries,
+// under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+// SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES.
+//
+//===----------------------------------------------------------------------===//
+
+#include <cuda/std/type_traits>
+#include <cuda/std/utility>
+
+#include <cuda/experimental/green_context.cuh>
+#include <cuda/experimental/stream.cuh>
+
+#include <testing.cuh>
+#include <utility.cuh>
+
+#if _CCCL_CTK_AT_LEAST(12, 5)
+C2H_TEST("Green context", "[green_context]")
+{
+  if (test::cuda_driver_version() < 12050)
+  {
+    SUCCEED("Driver is too old for green context tests");
+  }
+  else
+  {
+    INFO("Can create a green context");
+    {
+      {
+        [[maybe_unused]] cudax::green_context ctx(cuda::devices[0]);
+      }
+      {
+        cudax::green_context ctx(cuda::devices[0]);
+        auto handle     = ctx.release();
+        auto new_object = cudax::green_context::from_native_handle(handle);
+      }
+    }
+
+    INFO("Can create streams under green context");
+    {
+      cudax::green_context green_ctx_dev0(cuda::devices[0]);
+      cudax::stream stream_under_green_ctx(green_ctx_dev0);
+      REQUIRE(stream_under_green_ctx.device() == 0);
+      if (cuda::devices.size() > 1)
+      {
+        cudax::green_context green_ctx_dev1(cuda::devices[1]);
+        cudax::stream stream_dev1(green_ctx_dev1);
+        REQUIRE(stream_dev1.device() == 1);
+      }
+
+      INFO("Can create a side stream");
+      {
+        auto ldev1 = stream_under_green_ctx.logical_device();
+        REQUIRE(ldev1.kind() == cudax::logical_device::kinds::green_context);
+        cudax::stream side_stream(ldev1);
+        REQUIRE(side_stream.device() == 0);
+        auto ldev2 = side_stream.logical_device();
+        REQUIRE(ldev2.kind() == cudax::logical_device::kinds::green_context);
+        REQUIRE(ldev1 == ldev2);
+      }
+    }
+  }
+
+#  if _CCCL_CTK_AT_LEAST(13, 0)
+  if (test::cuda_driver_version() >= 13000)
+  {
+    INFO("Can get green context ID");
+    {
+      STATIC_REQUIRE(cuda::std::is_same_v<unsigned long long, cuda::std::underlying_type_t<cudax::green_context_id>>);
+      STATIC_REQUIRE(
+        cuda::std::is_same_v<cudax::green_context_id, decltype(cuda::std::declval<cudax::green_context>().id())>);
+
+      cudax::green_context ctx1{cuda::devices[0]};
+      cudax::green_context ctx2{cuda::devices[0]};
+
+      // Test that id() returns a valid ID
+      auto id1 = ctx1.id();
+      auto id2 = ctx2.id();
+
+      // Test that different contexts have different IDs
+#    if _CCCL_COMPILER(NVHPC, <, 25, 11)
+      REQUIRE(cuda::std::to_underlying(id1) != cuda::std::to_underlying(id2));
+#    else // ^^^ _CCCL_COMPILER(NVHPC, <, 25, 11) ^^^ / vvv !_CCCL_COMPILER(NVHPC, <, 25, 11) vvv
+      REQUIRE(id1 != id2);
+#    endif // ^^^ !_CCCL_COMPILER(NVHPC, <, 25, 11) ^^^
+
+      // Test that the same context returns the same ID when called multiple times
+#    if _CCCL_COMPILER(NVHPC, <, 25, 11)
+      REQUIRE(cuda::std::to_underlying(ctx1.id()) == cuda::std::to_underlying(id1));
+      REQUIRE(cuda::std::to_underlying(ctx2.id()) == cuda::std::to_underlying(id2));
+#    else // ^^^ _CCCL_COMPILER(NVHPC, <, 25, 11) ^^^ / vvv !_CCCL_COMPILER(NVHPC, <, 25, 11) vvv
+      REQUIRE(ctx1.id() == id1);
+      REQUIRE(ctx2.id() == id2);
+#    endif // ^^^ !_CCCL_COMPILER(NVHPC, <, 25, 11) ^^^
+    }
+  }
+#  endif // _CCCL_CTK_AT_LEAST(13, 0)
+}
+#else // ^^^ _CCCL_CTK_AT_LEAST(12, 5) ^^^ / vvv _CCCL_CTK_BELOW(12, 5) vvv
+// For some reason CI fails with empty test, add a dummy test case
+C2H_TEST("Dummy test case", "")
+{
+  REQUIRE(1 == 1);
+}
+#endif // ^^^ _CCCL_CTK_BELOW(12, 5) ^^^